private static IEnumerable <string> GenerateEmails(
SubstringsData substrings,
SubstringWithCountList domains,
SubstringWithCountList tlds,
LengthDistribution lengthDistribution,
SampleValuesGeneratorConfig.Result config)
{
var rand = new Random(Environment.TickCount);
return(Enumerable.Range(0, 100 * config.SamplesToPublish)
.Select(_ => GenerateEmail(substrings, domains, tlds, lengthDistribution, config, rand))
.Where(email => !string.IsNullOrEmpty(email))
.Take(config.SamplesToPublish));
}
private static string GenerateEmail(
SubstringsData substrings,
SubstringWithCountList domains,
SubstringWithCountList tlds,
LengthDistribution lengthDistribution,
SampleValuesGeneratorConfig.Result config,
Random rand)
{
// create local-part section
var str = GenerateString(substrings, lengthDistribution, minLength: 6, rand);
if (string.IsNullOrEmpty(str))
{
return(string.Empty);
}
var allParts = str.Split('@', StringSplitOptions.RemoveEmptyEntries);
var sb = new StringBuilder();
var partIndex = 0;
var pnext = 1;
while (partIndex < allParts.Length && rand.NextDouble() <= pnext)
{
sb.Append(allParts[partIndex]);
pnext /= 2;
partIndex++;
}
var localParts = sb.ToString()
.Split('.', StringSplitOptions.RemoveEmptyEntries)
.Where(s => (s.Length == 1 || s.Length > 3) && !BannedWords.Contains(s.ToUpperInvariant()));
var localPart = string.Join('.', localParts);
if (string.IsNullOrEmpty(localPart))
{
return(string.Empty);
}
if (domains.TotalCount > config.MinValuesForCategoricalSampling)
{
// if the number of distinct domains is big enough we select one from the extracted list
return(localPart + domains.GetRandomValue(rand));
}
// create domain section
sb.Clear();
while (partIndex < allParts.Length)
{
sb.Append(allParts[partIndex]);
partIndex++;
}
var domainParts = sb.ToString()
.Split('.', StringSplitOptions.RemoveEmptyEntries)
.Where(p => p.Length > 3 && !BannedWords.Contains(p.ToUpperInvariant()));
var domain = rand.NextDouble() > 0.15 ?
domainParts.Aggregate(string.Empty, (max, cur) => max.Length > cur.Length ? max : cur) :
string.Join('.', domainParts);
if (string.IsNullOrEmpty(domain) || domain.Length < 4)
{
return(string.Empty);
}
return(localPart + "@" + domain + tlds.GetRandomValue(rand));
}